Simmo Saan

That's quite significant, although I don't remember how large they were at the time (maybe I still have some laying around). There isn't much that can be done about it...

> This may also have some precision benefits in cases where a malloc is thread-local and the block never needs to be escaped, as it can be handled flow-sensitively in...

Somehow every unlocked CI job on master after merging this is failing with ``` Fatal error: exception Sys_error("run/config.json: No such file or diretory") ```

This was changed to 5 in #1234, so I suppose we can consider it done for now?

> However, if we do not manage to do all that before SV-COMP 2025, we could at least enable it with `1` for no-overflows. This should be a fairly simple...

#25890 removed python3-distutils depext from 4.13.0 by releasing 4.13.0-1. It remained removed in 4.13.0-2 and 4.13.0-3. However, 4.13.2 and subsequent releases have the python3-distutils depext back again, making them not...

> Produces a warning for the comparison (from `invariant` while trying to perform the HC4-Revise algorithm): It would seem like any actual warnings should be produced during forward evaluation, so...

> It would be possible to patch this but this would require equipping every operation with a `~suppress_ovwarn` argument as in #1406. I think that is quite ugly, so I...

Other than a clean general fix, did #1411 fix everything here? If so, then we can close this and open a separate issue about the cleanup/refactoring. **EDIT:** Ah no, #1408...

The new Apron is now published on opam (https://github.com/ocaml/opam-repository/pull/26174) and all versions of Goblint are incompatible (https://github.com/ocaml/opam-repository/pull/26199), so this fix is crucial for release. This is probably easiest done after...